How to properly cite your preprint while preparing the camera-ready paper

When updating a paper already shared on ArXiv for the final conference version, authors must carefully navigate how to reference their own work without undermining its originality or appearing self-serving. The challenge lies in acknowledging the preprint’s role—where it laid foundational groundwork—while ensuring the revised version maintains its distinct contributions.

Citing the preprint is not only acceptable but often necessary when the conference paper builds upon technical details or experimental setups that exceeded the original page limits. In fields like machine learning, ArXiv serves as the standard repository for such supplementary material. The key is to integrate the citation thoughtfully rather than as a standalone reference. For instance, a footnote on the first page clarifies the relationship:

\thanks{A preliminary version of this work appeared in [Preprint Citation/ArXiv ID].}

This approach avoids inflating citation metrics while ensuring transparency for reviewers and readers.

A more complex issue arises when subsequent works adopt or extend the original methodology. While these papers validate the preprint’s impact, their inclusion must not diminish the original contribution. The solution is to frame later works as extensions rather than competitors. Instead of phrasing like "Work X also uses our method to solve Y," authors should emphasize their method as the foundational baseline:

"Our proposed methodology has recently been adopted by several studies to address [specific problem] (e.g., [Citation]), demonstrating its versatility in [specific domain]."

This strategy reinforces the original work’s influence while positioning later research as applications of the core innovation.

Finally, when refining technical details—such as prompt engineering or architectural improvements—authors should ensure the specific novelty introduced in their paper remains clear. Even if later works build upon the framework, the original contribution should be highlighted as the essential foundation. For example:

"While [Paper B] extends our framework to include [new feature], our original work provides the fundamental [core mechanism] necessary for [primary task]."

This narrative approach transforms potential concerns about novelty into evidence of broader impact, demonstrating that the research has already shaped practical AI workflows.